Machine Learning

While machine learning sounds highly technical, an introduction to the statistical methods involved quickly brings it within reach. In this article, Toptal Freelance Software Engineer Vladyslav Millier explores basic supervised machine learning algorithms and scikit-learn, using them to predict survival rates for Titanic passengers.

TensorFlow is more than just a machine intelligence framework. It is packed with features and tools that make developing and debugging machine learning systems easier than ever. In this article, Toptal Freelance Software Engineer Dino Causevic gives us an overview of TensorFlow and some auxiliary libraries to debug, visualize, and tweak the models created with it.

TensorFlow makes implementing deep learning on a production scale a breeze. However, understanding its core mechanisms and how dataflow graphs work is an essential step in leveraging the tool’s power. In this article, Toptal Freelance Software Engineer Oliver Holloway demonstrates how TensorFlow works by first solving a general numerical problem and then a deep learning problem.

Generative adversarial networks, among the most important machine learning breakthroughs of recent times, allow you to generate useful data from random noise. Instead of training one neural network with millions of data points, you let two neural networks contest with each other to figure things out. In this article, Toptal Freelance Software Engineer Cody Nash gives us an overview of how GANs work and how this class of machine learning algorithms can be used to generate data in data-limited situations.

Many problems have optimal algorithms developed for them, while many others require us to randomly guess until we get a good answer. Even an optimal solution becomes slow and complex at a certain scale, at which point we can turn to natural processes to see how they reach acceptable results. In this article, Toptal Freelance Software Engineer Eugene Ossipov walks us through the basics of creating a Genetic Algorithm and gives us the knowledge to delve deeper into solving any problems using this approach.

As a language, R is strongly tied to data and is thus used mostly by statisticians and data scientists. Many who already use R for machine learning, though, are not aware that data munging can be done faster in R, meaning another tool is not required for that task. In this article, Freelance Software Engineer Jan Gorecki explores tabular data transformations and introduces us to one of the fastest open-source data wrangling tools available.

More than 60 percent of trading activities with different assets rely on automated trading and machine learning instead of human traders. Today, specialized programs based on particular algorithms and learned patterns automatically buy and sell assets in various markets, with a goal to achieve a positive return in the long run. In this article, Toptal Freelance Data Scientist Andrea Nalon explains how to predict, using machine learning and Python, which trade should be made next on the S&P 500 to get a positive gain.

Clustering algorithms are very important to unsupervised learning and are key elements of machine learning in general. These algorithms give meaning to data that are not labelled and help find structure in chaos. But not all clustering algorithms are created equal; each has its own pros and cons. In this article, Toptal Freelance Software Engineer Lovro Iliassich explores a heap of clustering algorithms, from the well known K-Means algorithm to the elegant, state-of-the-art Affinity Propagation technique.

Today, a massive amount of data is available in the form of networks or graphs. For example, the World Wide Web, with its web pages and hyperlinks, social networks, semantic networks, biological networks, citation networks for scientific literature, and so on. A tree is a special type of graph, and is naturally suited to represent many types of data. The analysis of trees is an important field in computer and data science. In this article, we will look at the analysis of the link structure in trees. In particular, we will focus on tree kernels, a method for comparing tree graphs to each other, allowing us to get quantifiable measurements of their similarities or differences. This an important process for many modern applications such as classification and data analysis.